Peter, Markus took these words out of my
mouth.
In general, to avoid confusion like
this, I suggest we use the following guidelines when writing: a) X3 snippets,
and b) XDI addresses:
1) In the X3 snippets, use one of the X3
formats documented at http://wiki.oasis-open.org/xdi/X3Format.
In other words, use either square brackets (standard X3) or multiple lines
(X3D) or both (X3W).
2) When writing XDI addresses that resolve
against graphs shown in X3, ONLY include the actual XRIs, and not literal
values, which are NOT part of the address (the address only includes the XRIs
needed to identify the target node(s)).
So, as Markus corrected himself, his
example should have been:
GRAPH – X3D FORMAT
=markus
+email!1
“”
OR GRAPH – X3W FORMAT
[=markus
[+email!1
[“”]
]
]
OR GRAPH – X3 FORMAT
[=markus[=email!1[“”]]]
XDI ADDRESS IN THIS GRAPH
=markus/+email!1
Hope this helps,
=Drummond
From:
[mailto:] On Behalf Of Markus Sabadello
Sent: Tuesday, January 29, 2008
11:05 AM
To: Peter Davis
Cc: Giovanni Bartolomeo;
; Drummond Reed
Subject: Re: [xdi] Re: again on
numbering..
Heya Peter,
Hmm maybe there's a misunderstanding.. When I wrote lines like this:
=markus/+email!1/""
..I didn't mean this as an identifier that points somewhere. It's a
very short way of writing down XDI data itself that's stored at some endpoint
(e.g. at =markus' XDI endpoint)
The same data in X3 would be [=markus[+email!1[""]]].
A POINTER to that data would look exactly as you propose I think,
e.g.
=markus/+email!1
So there's no literal in the identifier itself, only in the XDI
document the identifier points to.
Does that make any sense?
Markus
On 1/29/08, Peter
Davis <>
wrote:
A general observation wrt
the X3 graphs being proposed here:
I am seeing an increase in the discussions were predicate values are
being explicitly declared in the XRI productions. I think this
approach can be wrought with peril...
My concern revolves around the exposure of communication endpoints in
identifiers. This was a property that made XRI's to me very
compelling. I can share some concrete identifier (and a path into,
for example, to a predicate eg: xri://=peterd/+email) without
exposing the email literal. rather, a service endpoint which might
handle XDI (for example) requests can apply some set of security
policies before responding with some (if any) values. Out of
curiosity, why would i want to express the value of a predicate in
the identifier itself (rather than a pointer to it's value)?
Here i am seeing attributes of identity being regularly used in the
identifiers, and thus the privacy properties of the communication
endpoints are lost.
I realize that i have not been active on this list for some time, and
so i may be taking the use of the identifiers being expressed here
out of context, but if my interpretation of this discussion is
correct, i see what i think is a misuse of XRI as an address into my
attribute graph.
my .02
=peterd
On Jan 29, 2008, at 12:41 PM, Markus Sabadello wrote:
> Heya Giovanni,
>
> I think the idea of having a way to express ordered collections in
> XDI would be very useful...
>
> We came across a similar problem a few days ago.. How do we express
> multiple values for a predicate? For example, =markus could have
> multiple e-mail addresses. We came up with the following idea:
>
> =markus/+email!1/""
> =markus/+email!2/""
>
> But your pattern would work too:
>
> =markus+email/$num$1/""
> =markus+email/$num$2/""
>
> A third idea Drummond had was to use inner graphs, like this:
>
> =markus
> +email
> /
> $
>
$1
>
""
>
$2
>
""
>
> I like all three methods.. But we should probably decide on one of
> them.. Or just experiment for a while and see what works best...
>
> Markus
>
> On 1/29/08, Giovanni Bartolomeo <>
> wrote: Hello,
>
> some new inputs about this issue:
>
> >* Markus asked if $num is always used in conjuntion with $has
> predicate, or
> >if there would be other uses? Giovanni said it would also be useful
> >standalone.
>
> I think that $num could be useful also in accessing arrays, vectors
> and similar structures what we could call "ordered collections".
>
> Let's consider again this use case:
>
> =football.team/$has$num$11/+player
> @as.roma.calcio/$is$a/+football.team
>
> we could index each player using $num${x} as follow:
>
> @as.roma.calcio+player/$num$0/=curci
> ...
> @as.roma.calcio+player/$num$7/=derossi
> ...
> @as.roma.calcio+player/$num$10/=totti
>
> etc.
>
> Do you think this capability could be of help for XDI?
>
> BTW maybe we can avoid the prefix $num and leave only ${num}, e.g.
> $0, $5, $7 are valid XRIs which represent the number 0, 5, 7.
>
> What do you think?
>
> Bests,
> Giovanni
>
>
>
>
